# unum-cloud/UForm

Pocket-Sized Multimodal AI for content understanding and generation across multilingual texts, images, and 🔜 video, up to 5x faster than OpenAI CLIP and LLaVA 🖼️ & 🖋️

Repository: https://github.com/unum-cloud/UForm
Canonical: https://ross.abutalabs.com/products/uform
Homepage: https://unum-cloud.github.io/UForm
Language: Python
License: Apache-2.0
License Family: permissive
Topics: huggingface-transformers, language-vision, multimodal, pytorch, semantic-search, transformer, cross-attention, vector-search, bert, neural-network, pretrained-models, multi-lingual, clip, openai, openclip, contrastive-learning, representation-learning, clustering, image-search, llava
Last push: 2025-10-30T23:39:54+00:00

## Health v2 (maintenance only)
Score: 55/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 49, release rhythm 42, longevity 92
- inputs: {"age_days": 1289, "days_push": 307, "days_rel": 307, "gap_med": 73, "n_releases_24m": 6}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1244, forks 78 (observed 2026-08-28T04:04:06.750007+00:00)

## What it is
UForm is a compact multimodal AI library providing tiny image-text embedding models (64-768 dimensions, Matryoshka-style) and small generative chat models for image captioning and visual question answering. It ships pretrained transformers with ONNX, CoreML, and PyTorch support, deployable from servers to smartphones.

## Use cases
- compute image and text embeddings for semantic search
- search images with natural language queries
- caption images automatically
- build visual question answering
- run multimodal AI on mobile devices
- embed multilingual text for cross-lingual retrieval

## When to choose
- you need fast, tiny CLIP-like embeddings with multilingual support
- you want to deploy multimodal models on edge devices or smartphones
- you need quantization-friendly embeddings for vector search

## When to avoid
- you need state-of-the-art accuracy over speed and size
- you require video or long-document understanding today
- you need large-scale generative LLM capabilities

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, nlp, image-processing, llm-inference, rag, search-engine
- domain: machine-learning, artificial-intelligence, computer-vision, deep-learning
- platform: python, cross-platform
- tags: multimodal, embeddings, clip, image-captioning, vqa, onnx, matryoshka-embeddings, semantic-search, contrastive-learning, quantization, natural-language-processing, search, javascript, swift, gpu, mobile

## Member repositories
- unum-cloud/UForm (main) score 55

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:06.750007+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T05:08:17.071583+00:00, confidence not recorded.
  - readme: https://github.com/unum-cloud/UForm (fetched 2026-08-28T04:04:06.750007+00:00, sha 5de8a89b2845)
  - homepage: https://unum-cloud.github.io/UForm (fetched 2026-08-29T12:19:58.935120+00:00, sha 201427ae3109)
  - registry_pypi: https://pypi.org/pypi/uform/json (fetched 2026-08-29T12:19:58.937898+00:00, sha 15edf40abda3)
- Data as of 2026-08-30T08:39:29.467469+00:00.